Best Time to Send Cold Emails: How to Test It on Your Own List

A published send time describes the people in someone else’s data. Several things stop it from describing yours.

The audience is different. Founders, plant managers, finance leads and agency owners keep different days. A finding drawn from a mix of all of them describes none of them well.

Time zones get blended. If a study logs send times in the sender’s zone, “10am” might mean 10am for one recipient and 7pm for another. Unless the data was converted to each recipient’s local time, the hour in the headline is close to meaningless.

Volume gets mistaken for performance. Most teams send in the morning, so most replies arrive in the morning. A chart of replies by hour often shows when people sent, not when sending worked. The figure that matters is the rate per person contacted in each window.

Survivorship leans the story one way. Campaigns that went well get written up, along with whatever timing they happened to use. Campaigns that flopped at the same hour rarely make it into the post.

The same problems affect published reply rates, as the guide to cold email benchmarks sets out. The fix is the same too: measure your own buyers, with fixed definitions and a known sample size.

A lot of send-time advice is built on opens: send at the hour when most emails get opened. That measure has become unreliable.

Opens are usually tracked with a small image that loads when an email is viewed. Apple’s Mail Privacy Protection changes that for people who turn it on. According to Apple’s description of the feature, it downloads remote content in the background by default, regardless of whether the person engages with the email. Without it, senders can see when and how many times an email was opened. With it, an “open” can be recorded when the email arrives, whether or not anyone reads it.

For send-time work that is a serious problem. If an open is logged at delivery, the open time simply mirrors the send time. Every window looks like it works, and the ones you sent most look best.

Opens had a weakness even before that. An open shows that an email was displayed, not that it interested anyone. Some teams also turn off open tracking in cold outreach to keep messages plain.

Measure send time on positive replies and meetings booked, counted per person contacted. They are rarer than opens, so tests need more people, but they measure what you actually want. The guide to outreach metrics that matter gives the formulas.

Some timing effects are plausible enough to be worth testing. Treat each as a hypothesis about your buyers, not a fact.

• The recipient’s local time. An email that arrives during their working day may be read sooner than one that arrives overnight and sits under the morning’s mail. This is the first thing to get right: send in the prospect’s time zone, not yours.
• Workday versus weekend. A weekend email may face a quieter inbox, or it may be skimmed on a phone and forgotten by Monday. Working weeks and public holidays also differ by country, so “weekday” needs defining per market.
• Role patterns. A founder may clear email late at night. Someone in operations may be on a site floor all morning. A finance lead may be buried at month end. Your own replies will show whether any of this holds.
• The best day to send cold email. Monday may compete with the week’s backlog and Friday with the end of it. Or neither matters for your list. Day of week is a fair thing to test, one comparison at a time.

Notice what is missing: a single best hour for everyone. If your buyers sit across several regions or roles, the right answer may differ between segments. Those segments should come from your ideal customer profile, so a timing result feeds straight back into who you target.

A send-time test compares two windows on the same kind of people with everything else held still. It takes seven steps.

1. 1Pick one segment, for example operations leaders at manufacturers in one country. Mixing segments lets audience differences pass as timing effects.
2. 2Choose one variable and two windows, such as 8–10am versus 1–3pm in the recipient’s local time. Test hour or day, not both at once.
3. 3Split the segment at random. Shuffle the list, then alternate people between the two groups. Never split by company name, by list source or by who was added first.
4. 4Keep everything else the same: the same copy, the same follow-up steps and gaps, and the same mailboxes sending to both groups. Only the first email’s send window changes.
5. 5Run both windows on the same days, across at least two full weeks, so both groups meet the same holidays, news and busy weeks.
6. 6Fix a reply window, for example replies within 14 days of the first email, and count every person against it, including those who replied to a follow-up.
7. 7Write the result down as a dated table with counts beside every rate, whatever it shows.

One row per window is enough. An illustrative example of the template, filled in:

Add a notes line under the table for anything unusual during the test, such as a paused mailbox, a public holiday in one region or a change to the offer.

The afternoon window in the example looks better: 4.4% against 3.0%. Before acting on it, estimate how big a gap chance alone could produce. A standard rough approximation for the difference between two rates:

Margin ≈ 2 × √( rate A × (1 − rate A) ÷ people A + rate B × (1 − rate B) ÷ people B )

For the example, that comes to about 2.4 percentage points. The observed gap is 1.4 points, well inside the margin. On this sample, the two windows can’t be told apart. The meetings, four against six, are even less conclusive.

How many people would it take? If the true rates really were 3% and 4.5%, you would need roughly 1,300 people in each group before a gap that size cleared this rough margin. A formal power calculation, which also guards against missing a real difference, asks for about twice that. Smaller differences need far more.

That has a practical meaning for small teams. Many will never contact enough people in one segment to prove a modest timing effect. That is useful to know. It means timing is not where your next improvement will come from.

• The gap is larger than the margin, and holds when you repeat the test
• The direction is the same for positive replies and for meetings
• Neither window was hit by a holiday, outage or paused mailbox
• Both groups came from the same segment, split at random

If a result fails these checks, choose the window that is easier to run and move on. A tie is a real finding: it says send time is not a lever for this segment.

Writing to the wrong person at the perfect hour still gets no reply. Writing to the right person about a problem they have this quarter tends to work at most reasonable hours. Of the three, who you contact and what you say usually move results far more than when the email lands.

Chasing a perfect hour can also cost you. If every email goes out at 9:00 sharp, a mailbox sends its whole day in a few minutes. That burst looks less like a person writing email and more like a bulk send, and a bad batch of addresses or a run of complaints hits all at once. Spread sending through the working day, inside sensible daily limits, as the guide to email deliverability for cold outreach recommends.

A sensible default needs no study: send during the recipient’s working hours, on their working days, spread across the day. Then put your testing effort into segments and messages, and run a send-time test only when a segment is large enough to answer the question.